Ml Paper Writing

This handles the full workflow of writing ML papers for NeurIPS, ICML, ICLR, and similar venues, starting from a research repo with code and results. It's built around one critical rule: never hallucinate citations. Every reference must be fetched programmatically through APIs like Semantic Scholar or Exa MCP, with explicit placeholders for anything unverified. The philosophy is proactive collaboration. Claude explores your repo, understands the contribution, and delivers a complete first draft rather than blocking on every section. You then iterate based on feedback. It includes LaTeX templates, conference-specific checklists, and citation verification workflows. Most useful when you have experimental results but need to shape them into publication-ready prose without inventing fake references.